The epidemiology of breast cancer suggests a central etiologic role for premenopausal hormones, yet limited data exist to indicate which specific hormones are of greatest importance and at what levels risk is increased. Recently, we observed significant positive associations between incident premenopausal breast cancer and plasma levels of the sex steroids estradiol, androstenedione, testosterone, and dehydroepiandrosterone sulfate. In this application, we propose to extend these initial findings - for example, we will better determine the dose-response relation between plasma levels and disease and assess associations by tumor characteristics (e.g., ER+/PR+ vs. ER-/PR-). Further, we plan to evaluate urinary estrogen metabolites using a newly developed assay. We also propose several new steroid- hormone related aims, including the assessment of plasma enterolactone (a phytoestrogen), 25-OH vitamin D, and genetic variation in the vitamin D pathway. Finally, to better understand the contribution of these factors to breast cancer risk, we propose to evaluate their addition to both the Gail and Rosner statistical risk prediction models. In this study, we will utilize blood and urine samples collected in 1996-99 from 29,611 participants in the Nurses'Health Study II (NHSII) who were 32 to 54 years of age at collection. 18,521 of these samples were timed according to the woman's menstrual cycle;for these women, both early follicular and midluteal blood samples are available. Samples have been stored at -130oC or colder in liquid nitrogen freezers since collection. We will assay samples from women who were diagnosed with breast cancer after blood collection and matched controls who remained disease-free (i.e., a nested case-control design), thus efficiently utilizing these prospectively collected samples. By 2009, we expect to confirm 766 incident cases of breast cancer, 435 of which will have occurred among those with timed samples. The NHSII is one of a few cohort studies that provides an ideal context in which to investigate these relationships. Few other studies worldwide have a large number of blood and urine samples from predominantly premenopausal women. Also, follow-up in this cohort remains high (98% in the blood cohort) and a wealth of data on reproductive factors, diet, physical activity, and other variables has been collected since 1989 and continues to be collected (CA50385;Willett, PI). Hence, our proposed study should contribute substantially to our understanding of the relationships between hormones and breast cancer in younger women. Further, confirmation of protective associations for vitamin D or enterolactone, factors amenable to change through lifestyle choices, would have important public health implications.